TWILITE: A recommendation system for Twitter using a probabilistic model based on latent Dirichlet allocation
作者：
Highlights：
• We propose a probabilistic model by generalizing LDA to capture the behaviors of users in Twitter.
• A variational EM algorithm is presented for our model.
• We develop recommendation algorithms using the parameters estimated by the inference algorithm.
• We did performance study to show the effectiveness of our model and recommendation algorithms.
• Our recommendation algorithms are much more accurate than traditional algorithms in experiments.
摘要
Highlights•We propose a probabilistic model by generalizing LDA to capture the behaviors of users in Twitter.•A variational EM algorithm is presented for our model.•We develop recommendation algorithms using the parameters estimated by the inference algorithm.•We did performance study to show the effectiveness of our model and recommendation algorithms.•Our recommendation algorithms are much more accurate than traditional algorithms in experiments.
论文关键词：Twitter,Recommendation system,Probabilistic model,Collaborative filtering,LDA,Matrix factorization
论文评审过程：Received 7 November 2013, Revised 24 November 2013, Accepted 26 November 2013, Available online 4 December 2013.
论文官网地址：https://doi.org/10.1016/j.is.2013.11.003
